Alpha Testing
Alpha Testing is where you become the strategist. Write a trading strategy in plain English, pick an AI model to execute it, and watch the results unfold in real time. No coding required; just your ideas, translated into action by state-of-the-art language models.
How It Works
1. Write Your Strategy Prompt
Describe your trading approach in natural language. There is no special syntax or programming language to learn. Write the way you would explain your strategy to a knowledgeable human trader.
Example prompts:
"Focus on large-cap coins. Buy when RSI drops below 30 and sell when it crosses above 70. Never allocate more than 20% of the portfolio to a single asset."
"Trade only BTC and ETH. Follow momentum: buy breakouts above the 24-hour high, use a 3% trailing stop. Be aggressive in uptrends and sit in cash during choppy markets."
"Diversify across at least 5 assets. Rebalance weekly. Overweight assets showing increasing volume and underweight assets with declining on-chain activity."
The more specific your prompt, the more precisely the AI model will follow your intent. You can reference technical indicators, market conditions, risk parameters, position sizing rules, and any other aspect of your trading philosophy.
2. Select an AI Model
Choose from the available AI models. Each model has distinct strengths: some excel at nuanced reasoning, others at speed, and others at handling large amounts of context. See Supported AI Models for a full breakdown.
You can also run the same strategy across multiple models simultaneously to compare how different AIs interpret and execute identical instructions.
3. Set Your Investment Amount
Define how much capital you want to allocate to this strategy. On mainnet, this is real money and real risk -- you can lose funds. On Sepolia testnet, trading uses testnet tokens with no monetary value, making it a risk-free environment for experimentation.
4. Launch and Monitor
Once launched, the AI model begins trading according to your prompt. You can monitor:
- Live portfolio value: Track your strategy's equity curve in real time.
- Trade log: See every buy and sell the AI executes, with prices and timestamps.
- Performance metrics: Review returns, win rate, drawdown, and other key statistics.
- AI reasoning: Understand why the model made each trading decision.
Multi-Model Testing
One of the most powerful features of Alpha Testing is the ability to test the same prompt across multiple AI models. This reveals how different models interpret identical instructions:
- Does one model trade more conservatively than another?
- Which model better understands your risk parameters?
- How do premium-tier models compare to economy-tier models on the same strategy?
Multi-model testing turns every experiment into a controlled comparison, helping you find the best model for your specific approach.
Community Features
Alpha Testing is not a solo experience. The platform includes several community-driven features:
View Other Users' Trades
Browse trades from other community members to discover strategies you might not have considered. See which prompts and models other users are experimenting with, and learn from their results.
Leaderboards
Community leaderboards rank strategies by performance. See which combinations of prompts and models are producing the strongest returns, the most consistent results, or the best risk-adjusted performance.
Share Your Results
Had a successful run? Share your strategy results with the community. Comparing notes with other users accelerates learning and helps the entire community discover what works.
Writing Effective Prompts
Your prompt is the single most important input. Here are some guidelines:
Be Specific About Risk
Vague risk instructions lead to unpredictable behavior. Instead of "be careful," try:
"Never risk more than 5% of total portfolio value on a single trade. Set a stop-loss at 3% below entry for every position. Move to 100% cash if the portfolio drops more than 10% from its peak."
Define Your Universe
Tell the AI which assets to consider:
"Only trade BTC, ETH, SOL, and LINK. Ignore all other assets."
Or let it choose with guidelines:
"Trade any asset in the top 10 by market cap. Avoid meme coins entirely."
Specify Timeframes
"Make decisions based on 4-hour candles. Do not react to movements on timeframes shorter than 1 hour."
State Your Goal
"Maximize absolute returns over a 7-day period."
"Minimize drawdown while targeting 2% weekly returns."
"Beat a simple buy-and-hold Bitcoin strategy."
Current Limitations
Alpha Testing is live but still in its early stages. Keep the following in mind:
Network Differences
Alpha Testing is available on both Ethereum mainnet and Sepolia testnet. On mainnet, you trade with real funds and face real market conditions including slippage, true liquidity depth, and actual transaction costs. On Sepolia testnet, you trade with testnet tokens that have no monetary value, providing a risk-free environment for experimentation. Note that testnet results may not perfectly reflect mainnet conditions.
Experimental AI Behavior
Language models are powerful but imperfect. They may occasionally misinterpret prompts, make unexpected trades, or exhibit inconsistent behavior across sessions. Treat all results as experimental.
Strategy Iteration
AI trading is not a "set and forget" activity. The best results come from iterating on your prompts: testing, observing, refining, and retesting. Expect to go through several versions before finding a strategy that performs the way you envision.
No Financial Guarantees
Even strategies that perform well in testing or over a given period may not continue to produce similar results. AI models do not have any inherent ability to predict future prices. Their value lies in consistent, rule-based execution, not in guaranteed returns. Past performance on testnet or mainnet does not guarantee future results.
Getting Started
- Connect your wallet to the platform.
- Navigate to the Alpha Testing section.
- Write your first strategy prompt.
- Select an AI model (or multiple models).
- Set your investment amount.
- Launch and watch the AI trade.
Start simple. A straightforward strategy like "Buy BTC when it drops 5% in 24 hours, sell when it recovers 3%" is a great way to learn how the system works before designing more complex approaches.
What's Next
Alpha Testing is the foundation of LLMTrader's trading experience. With mainnet now live, the strategies and insights you develop can be deployed with real capital immediately. You can also continue to use Sepolia testnet to refine your approach risk-free before committing real funds.
See Supported AI Models to learn about the models available for your strategies, or browse Supported Assets to see which cryptocurrencies you can trade.
Risk Warning: Mainnet Alpha Testing uses real funds. You can lose money. Losses on mainnet are real and irreversible. Testnet Alpha Testing uses tokens with no monetary value and carries no financial risk. Regardless of network, all strategies are experimental and should not be assumed to produce consistent results. Cryptocurrency trading is inherently risky. AI-generated trading decisions are not financial advice. Always conduct your own research and never invest more than you can afford to lose.